KR20190138733A - Determination of Health Markers Using Portable Devices - Google Patents

Abstract

 Processing of the PPG data using the device includes: determining a quality estimate for a segment of the PPG data; And in response to determining that the quality estimate exceeds a quality threshold, filtering the segment of the PPG data based on an estimate of the periodicity of the segment of the PPG data. Health markers can be determined for the segment of the PPG data. The health marker may be validated based on previous determination of the health marker from PPG data. Based on the validation, the health indicator can be output.

Description

Determination of Health Markers Using Portable Devices

The present disclosure relates to determining health markers for a user, and more particularly, to determining such health markers using a portable device.

The popularity of portable devices has increased significantly over the last few years. Examples of portable devices include mobile phones and tablet computers. Another category of portable devices is "wearable devices." An example of a wearable device is a "smart watch." Many portable devices have a variety of built-in sensors. Some portable devices also have communication ports and / or transceivers that allow the devices to couple with one or more external sensors.

One type of sensor that is typically included or coupled to a portable device is a photoplethysmogram (PPG) sensor. PPG sensors are optical sensors capable of performing volumetric measurements of organs. For example, a PPG sensor can measure the modulation of blood flow in the human body in response to a beat-to-beat ejection of blood from the heart to the aorta.

Portable devices equipped with PPG sensors can generate information that can be used to assess a user's health. However, the data generated by the PPG sensor is very sensitive to noise. For example, small movements and / or movements in a portable device relative to a user are often sufficient to cause significant noise in the data generated by the PPG sensor. This noise consequently makes the data virtually unavailable.

The accompanying drawings show one or more embodiments. However, the accompanying drawings should not be understood as limiting the invention only to the embodiments shown. Various aspects and advantages will become apparent upon review of the following detailed description with reference to the drawings.
1 illustrates an example architecture for an apparatus.
2 illustrates an example method of processing photoplethysmogram (PPG) data.
3 illustrates an example method of quality estimation for PPG data.
4 illustrates an example method of filtering and / or correction for PPG data.
5 illustrates an example method of validating a health marker.

In one or more embodiments, a method includes using a processor in response to determining a quality estimate for a segment of PPG data, and in response to determining that the quality estimate exceeds a quality threshold. And filtering the segment of the PPG data based on an estimate of the periodicity of the segment of the PPG data. The method includes using the processor to determine a health marker for the segment of the PPG data, and to determine the health marker based on prior determination of the health marker from the PPG data. Validation may be included. The method may include outputting the health indicator in response to the validation.

In one or more embodiments, an apparatus includes a memory configured to store instructions and a processor coupled to the memory. The processor is configured to initiate operations in response to the execution of the instructions. The operations may include determining a quality estimate for a segment of PPG data, and in response to determining that the quality estimate exceeds a quality threshold, determining the quality of the PPG data based on an estimate of the periodicity of the segment of the PPG data. Filtering the segment.

The operations may include determining a health indicator for the segment of the PPG data, and validating the health indicator based on a previous determination of the health marker from the PPG data. The operations may also include outputting the health indicator in response to the validation.

In one or more embodiments, a computer program product may include a computer readable storage medium having stored thereon program code. The program code may be executed by the processor to perform the operations. The operations may include determining a quality estimate for a segment of PPG data, and in response to determining that the quality estimate exceeds a quality threshold, determining the quality of the PPG data based on an estimate of the periodicity of the segment of the PPG data. Filtering the segment.

The operations may include determining a health marker for the segment of the PPG data, and validating the health marker based on a previous determination of the health marker from the PPG data. The operations may also include outputting the health indicator in response to the validation.

This summary section is provided only to introduce certain concepts, and not to identify any key or essential features of the claimed subject matter. Many other features and embodiments of the present invention will become apparent from the accompanying drawings and the following detailed description.

Although the present disclosure concludes with the claims defining novel features, it is believed that the various features described herein will be better understood upon consideration of the description in conjunction with the drawings. The process (s), machine (s), article of manufacture (s) and any variations thereof described herein are provided for purposes of illustration. Any specific structural and functional details described are not to be construed as limiting, and are skilled in the art to which the present invention pertains in various ways as a basis for the claims and in virtually any appropriately detailed structure. It should be interpreted as a representative basis for informing one. In addition, the terms and phrases used within the present disclosure are not intended to be limiting, but rather to provide an understandable description of the described features.

The present disclosure relates to determining health markers for users, and more particularly, to determining such health markers using portable devices. Reliable determination of health markers for users requires accurate measurement and analysis of photoplethysmogram (PPG). As defined within the present disclosure, the term "PPG data" refers to data generated by a PPG sensor that is in a portable device (eg, is part of a portable device) or coupled to a portable device. Examples of health markers that can be determined from the PPG data may include, but are not limited to, heart rate, heart rate variability (HRV), stress, blood pressure, respiration, and arterial tone. By accurately evaluating health markers from PPG data, portable devices can be used to reliably manage various medical conditions. For example, portable devices can be used to manage diabetes, high blood pressure, cardiovascular diseases, pulmonary diseases, mood disorders, and substance-abuse. In addition, accurate assessment of health markers also enables portable devices to more accurately assess the quality of life of individuals.

The high sensitivity of PPG sensors to noise often results in wide variability in the generated PPG data. For example, in the case of wearable devices with PPG sensors, the generated PPG data is very sensitive to artifacts induced by the movement of the portable devices relative to the individual wearing the wearable devices. Even small movements can cause large variability in the resulting PPG data, significantly limiting the usefulness of portable devices for monitoring user health.

According to the inventive arrangements described in the present disclosure, a portable device with a PPG sensor can generate PPG data. The portable device may distinguish different segments of PPG data from each other based on a quality assessment. The portable device may identify segments of lower quality PPG data that are more likely to result in inaccurate prediction or estimation of one or more health markers.

In certain embodiments, the portable device may discard segments of PPG data determined to be of lower quality so that such segments are not used to predict health indicator (s) for the user. The portable device may also generate notifications that indicate insufficiency of the PPG data. The notifications may indicate that adjustment of the portable device may be needed, for example, to obtain higher quality PPG data that increases the accuracy of the determined health markers.

In one or more embodiments, the portable device can apply multidimensional analysis to the PPG data. Each dimension of the multidimensional analysis may evaluate different aspects of the PPG data that are not evaluated by any other dimension of the multidimensional analysis. The portable device may determine that certain segments of PPG data have insufficient quality based on one or any combination of two or more of the dimensions of the multidimensional analysis being performed. By discarding segments of low quality PPG data, the portable device can avoid generating incorrect and / or misleading health indicators.

In one or more embodiments, the portable device can analyze the PPG data in near real time. In performing the near real-time processing of the PPG data, the portable device may be configured with a current or more recently determined health marker (s) and previously determined (eg, history) to validate the accuracy of the currently determined health markers. Historical health marker (s) can be compared. Quasi-real time processing, described herein, facilitates accuracy improvement over other techniques that attempt to perform real time prediction of health markers. In certain embodiments, the portable device may perform additional validation of the health markers in a post hoc manner. The validation performed may include correction of the PPG data.

Further aspects of the configurations of the present invention are described in more detail below with reference to the drawings. For simplicity and clarity of illustration, the elements shown in the figures are not necessarily drawn to scale. For example, the dimensions of some of the elements may be exaggerated relative to other elements for clarity. Also, where considered appropriate, reference numerals are repeated among the figures to indicate corresponding, similar, or identical features.

1 illustrates an example architecture 100 for an apparatus for use with one or more embodiments described within the present disclosure. The architecture 100 may be used to implement a data processing system, communication device, or other system suitable for storing and / or executing program code. In certain embodiments, the architecture 100 can be used to implement a portable device. For example, the architecture 100 may be used to implement wearable devices such as mobile phones, tablet computers, portable computers (eg, laptop computers), smart watches worn by users, or other fitness devices.

In the example of FIG. 1, the architecture 100 includes at least one processor 105. The processor 105 is coupled to the memory 110 through an interface circuit 115. The architecture 100 stores computer readable instructions (also called "program code") in the memory 110. The memory 110 is an example of a computer readable storage medium. The processor 105 executes the program code accessed from the memory 110 through the interface circuit 115.

The memory 110 includes one or more physical memory devices, such as, for example, local memory 120 and mass storage device 125. The local memory 120 is implemented as one or more non-permanent memory devices used in the actual execution of the program code. Examples of the local memory 120 include random access memory (RAM) and / or any of a variety of types of RAM (eg, static RAM, suitable for use by a processor in executing program code). Dynamic RAM). The mass storage device 125 is implemented as one or more persistent data storage devices. Examples of the mass storage device 125 include a hard disk drive (HDD), a solid state drive (SSD), flash memory, read-only memory (ROM), an erasable program. Erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EPROM), or other suitable memory. The architecture 100 may also include one or more cache memories (not shown) that provide for temporary storage of at least some program code to reduce the number of times program code must be retrieved from mass storage when executed. Can be.

Examples of the interface circuit 115 may include, but are not limited to, input / output (I / O) subsystems, I / O interfaces, bus systems, and memory interfaces. For example, the interface circuit 115 may be a combination of various bus structures and / or bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus. It can be implemented as any of these.

In one or more embodiments, the processor 105 and / or the interface circuit 115 are implemented as separate components. In one or more embodiments, the processor 105, the memory 110, and / or the interface circuit 115 are integrated into one or more integrated circuits. The various components in the architecture 100 may be coupled by, for example, one or more communication buses or signal lines (eg, interconnects and / or wires). have. In certain embodiments, the memory 110 is coupled to the interface circuit 115 through a memory interface, eg, a memory controller (not shown).

The architecture 100 may include a display 135. In certain embodiments, the display 135 is implemented as a touch-sensitive or touchscreen display capable of receiving touch input from a user. Touch-sensitive displays and / or touch-sensitive pads use any of a variety of available touch-sensitive technologies to detect contacts, movements, gestures, and breaks in contact. can do. Examples of touch sensitive technologies include capacitive, resistive, infrared, and surface acoustic wave technologies, and for determining one or more contact points with a touch sensitive display and / or device. Other proximity sensor arrays or other elements are included, but are not limited to these.

The architecture 100 may include a camera subsystem 140. The camera subsystem 140 may be coupled to the interface circuit 115 directly or via a suitable input / output (I / O) controller. The camera subsystem 140 may be coupled to the optical sensor 142. The optical sensor 142 may be implemented using any of a variety of techniques. Examples of the optical sensor 142 may include, but are not limited to, a charge coupled device (CCD) or a complementary metal-oxide semiconductor (CMOS) optical sensor. The camera subsystem 140 and the optical sensor 142 may perform camera functions such as image recording and / or video recording.

The architecture 100 may include an audio subsystem 145. The audio subsystem 145 may be coupled to the interface circuit 115 directly or via a suitable input / output (I / O) controller. The audio subsystem 145 can be coupled to a speaker 146 and a microphone 148 to facilitate voice support functions, such as voice recognition, voice replication, digital recording, and phone functions.

The architecture 100 may include one or more wireless communication subsystems 150. Each of the wireless communication subsystem (s) 150 may be coupled to the interface circuit 115 directly or via a suitable I / O controller (not shown). Each of the wireless communication subsystem (s) 150 can facilitate communication functions. Examples of the wireless communication subsystems 150 may include, but are not limited to, radio frequency receivers and transmitters, and optical (eg, infrared) receivers and transmitters. The specific design and implementation of the wireless communication subsystems 150 may depend on the specific type of architecture 100 being implemented and / or the communication network (s) on which the architecture 100 is intended to operate. .

As illustrative and non-limiting examples, the wireless communication subsystem (s) 150 may comprise one or more mobile networks (eg, GSM, GPRS, EDGE), a WiMax network, a WiFi network, short-range. range can be designed to operate over a wireless network (eg, a Bluetooth® network), NFC, and / or any combination thereof. The wireless communication subsystem (s) 150 may implement hosting protocols such that the architecture 100 can be configured as a base station for other wireless devices.

The architecture 100 may include one or more sensors 155. Each of the sensors 155 may be coupled to the interface circuit 115 directly or via a suitable I / O controller (not shown). Examples of the sensors 155 that may be included in the architecture 100 include facilitating orientation, lighting, and proximity functions, respectively, of a device using the architecture 100. Motion sensors, light sensors, and proximity sensors are included, but are not limited to these.

Other examples of the sensors 155 include location sensors (eg, GPS receivers and / or processors) capable of providing geographic location sensor data, used to determine the direction of magnetic north for direction navigation. Electronic magnetometer (e.g., an integrated circuit chip) capable of providing sensor data that can be generated, and data representing in three dimensions a change in speed and direction of motion of a device using the architecture 100 An accelerometer capable of providing an altimeter, and an altimeter (eg, integrated circuit) capable of providing data indicative of altitude, but are not limited thereto.

The architecture 100 may include a PPG sensor 160. The PPG sensor 160 may be coupled to the interface circuit 115 directly or via a suitable I / O controller (not shown). The PPG sensor 160 may be implemented as an optical sensor. The PPG sensor 160 may generate PPG, for example, PPG data for the user. PPG is an optically obtained plethysmogram. In general, PPG is a volumetric measure of an organ. In one or more embodiments, the PPG sensor 160 is implemented as a pulse oxymeter that determines the amount of oxygen carried in the blood by measuring the change in light absorption by shining light on the skin.

In certain embodiments, the PPG sensor 160 is a multi-channel PPG sensor (eg, having light emitting diodes capable of generating light of different wavelengths). The PPG sensor 160 may be used to determine one or more health markers and / or to determine one or more surrogate markers of one or more health markers. In one or more embodiments, health markers may be referred to as "vital signs."

Within the present disclosure, the term "PPG data" is used to mean data generated by the PPG sensor and output from the PPG sensor as described above. Within the present disclosure, the terms "PPG data" and "PPG signal" may sometimes be used interchangeably. It will be appreciated that the PPG signal (s) are specified directly by the PPG data.

The architecture 100 may further include one or more input / output (I / O) devices 165 coupled to the interface circuit 115. The I / O devices 165 may be coupled to the architecture 100, for example to the interface circuit 115, either directly or through an intervening I / O controller (not shown). Examples of the I / O devices 165 include a track pad, keyboard, display device, pointing device, one or more communication ports (eg, Universal Serial Bus (USB) ports). ), Network adapters, and buttons or other physical controls.

The network adapter allows the architecture 100 to be coupled to other systems, computer systems, remote printers, and / or remote storage devices through intervening private or public networks. Note refers to the circuit. Modems, cable modems, Ethernet interfaces, and wireless transceivers that are not part of the wireless communication subsystem (s) 150 are examples of different types of network adapters that may be used with the architecture 100. One or more of the I / O devices 165 may be adapted to control one or more or all of the sensors 150 and / or the functions of one or more of the wireless communication subsystem (s) 150. Can be.

The memory 110 stores program codes. Examples of program code include, but are not limited to, routines, programs, objects, components, logic, and other data structures. For illustrative purposes, the memory 110 stores an operating system 170 and application (s) 175. The applications 175 may include, for example, a PPG signal processing application.

In one or more embodiments, the PPG signal processing application, when executed, causes a device implemented using the architecture 100 and / or is in communication with the device implemented using the architecture 100. Other devices that may be linked to may perform various operations described herein. The memory 110 may also include data used by the operating system 170, data used by the application (s) 175, data received from user inputs, and the sensor (s) 155. One or more or all of and / or data generated by the PPG sensor 160, data received and / or generated by the camera subsystem 140, received and / or generated by the audio subsystem 145. Data may be stored, regardless of the data received and / or the data received by the I / O devices 165.

In one aspect, the operating system 170 and the application (s) 175, implemented in the form of executable program code, are executed by the architecture 100, more specifically by the processor 105, It is performed to perform the above described operations within the present disclosure. As such, the operating system 170 and the application (s) 175 may be considered an integrated part of the architecture 100. In addition, any data and / or program code generated, and / or operated by, the architecture 100 (eg, the processor 105) may be employed when employed as part of the architecture 100. It should be understood that these are functional data structures that impart functionality.

The memory 110 may also store additional program code. Examples of additional program code include: instructions to facilitate communication with one or more additional devices, one or more computers, and / or one or more servers; Graphical user interface (GUI) and / or UI processing; Sensor related processes and functions; Telephone related processes and functions; Electronic-messaging related processes and functions; Web browsing related processes and functions; Media processing related processes and functions; GPS and navigation related processes and functions; Security functions; And camera related processes and functions, including but not limited to web camera and / or web video functions.

The architecture 100 may further include a power source (not shown). The power source can power various elements of the architecture 100. In one embodiment, the power supply is implemented as one or more batteries. The batteries may be implemented using any of a variety of known battery technologies, whether disposable (eg, replaceable) or rechargeable. In another embodiment, the power source is configured to draw power from an external power source to supply power (eg, direct current (DC) power) to the elements of the device. In the case of a rechargeable battery, the power source may further include a circuit capable of charging the battery or batteries when coupled to an external power source.

The architecture 100 is provided for purposes of illustration and not of limitation. An apparatus and / or system configured to perform the operations described herein may use an architecture that is different from the architecture 100 of FIG. 1. Such other architecture may be a simplified version of the architecture 100, including a memory capable of storing instructions and a processor capable of executing the instructions.

In this regard, the architecture 100 may include fewer components than shown or additional components not shown in FIG. 1, depending on the particular type of device being implemented. In addition, the particular operating system and / or application (s) included may vary depending on the device type, such as types of I / O devices 165 included. In addition, one or more of the illustrated components may be incorporated into or otherwise form part of another component. For example, the processor may include at least some memory.

In one or more embodiments, an apparatus implemented using the same or similar architecture as the architecture of FIG. 1 may perform quasi-real time processing of PPG data. In certain embodiments, the apparatus can implement a cascaded processing technique, wherein each different processing step can produce a more refined representation of the PPG data.

For example, each step may represent one or more dimensions of the multi-dimensional analysis being performed and pass the generated representation (s) from cascade to the next step. Some of the calculations performed in the previous steps may be used in subsequent steps or steps. In certain embodiments, the cascaded approach employs orthogonal processing techniques within various steps. Orthogonality of the techniques means that each step operates in different and / or independent aspects of the PPG data.

In one or more embodiments, the near real time analysis performed by the device facilitates the assessment of the health markers against the baseline of certain health markers such as heart rate. For example, certain health markers are Autonomic Nervous System (ANS) measurements that can only be evaluated relative to baseline.

As an illustrative and non-limiting example, the near real time analysis allows the device to compare a health indicator, such as the user's heart rate, with the user's heart rate at any time in the past. For example, the device may compare the current heart rate determination with that determined a few seconds ago (eg, 10, 20, 30, 40, 50, or 60 seconds ago). Through this comparison, the device may determine how the health indicator has changed for the user and / or validate the health indicator (s).

2 illustrates an example method 200 of processing PPG data. The method 200 may be performed by a portable device having the same or similar architecture as the architecture described with reference to FIG. 1. The method 200 may be implemented as a near real time processing technique. In one or more embodiments, the method 200 may be performed segment by segment. A segment refers to a time window defined for PPG data. The window may be defined as PPG data within a given time span. The time interval may be one or more seconds (eg, 5, 10, 20, 30, 40, 50, or 60 seconds). In another example, the window may be defined as a predetermined number of PPG data samples. In certain embodiments, the segment of PPG data can be a sliding window that is moved across the PPG data.

The method 200 may begin with the device receiving PPG data from a PPG sensor. In general, the method 200 describes the processing of a particular segment of PPG data. It will be appreciated that as the window is moved across PPG data in time, the method 200 may be repeated to continue processing additional segments of the PPG data.

At block 205, the apparatus determines a quality estimate for the segment of the PPG data. In one or more embodiments, the apparatus can determine a signal-to-noise ratio (SNR) for the PPG data. The SNR may be determined for each segment. The SNR measurement allows the device to identify segments of PPG data that lack a PPG signal. Additional aspects of the block 205 are described in more detail with reference to FIG. 3 herein.

At block 210, the device determines whether the segment of PPG data is of low quality. For example, the device may compare the quality estimate of the segment of PPG data at block 205 with a quality threshold. The quality estimate may be such predetermined SNR, in which the PPG data is considered to be of little or no value below the predetermined SNR.

For example, for segments of PPG data having a quality estimate that does not exceed the quality threshold, any health markers determined or estimated using such data are unreliable. In response to determining that the segment of PPG data is of low quality (eg, the quality estimate does not exceed the quality threshold), the method 200 continues to block 215. In response to determining that the segment of PPG data is not low in quality (eg, the quality estimate exceeds the quality threshold), the method 200 continues to block 220.

At block 215, the device rejects the segment of PPG data. For example, the device may delete the segment having a quality level that does not exceed the quality threshold. In certain embodiments, in rejecting the segment, the device does not attempt to determine any health indicator using the segment of PPG data. The device may also provide a notification indicating that the PPG data segment is not suitable for use in determining health markers.

In one or more embodiments, the notification may provide instructions to the user to adjust the device to obtain higher quality PPG data later. As mentioned, although not shown in FIG. 2, the method 200 may be newly initiated to process additional PPG data segments of PPG data.

In block 220, the device corrects and / or filters the PPG data segment. In one or more embodiments, the apparatus may perform a calibration process on the segment that includes de-noising from the PPG data segment. The apparatus may also perform filtering on the segment of the PPG data. In certain embodiments, the apparatus performs a dynamic filtering process wherein the particular filter applied is selected from a plurality of different available filters based on one or more attributes of the PPG data segment itself. Additional aspects of the block 220 are described in more detail with reference to FIG. 4 herein. In one or more embodiments, the device selects the particular filter applied based on the periodicity of the PPG data segment.

At block 225, the device optionally determines the pulsatile amplitude of the PPG data segment. In order to distinguish PPG data collected from organic objects with a detectable heartbeat and PPG data collected from organic objects without a detectable heartbeat, the apparatus is adapted to variance of the PPG data. Alternatively, a running estimate of covariance may be performed. If the PPG sensor is a single wavelength sensor (eg, not multi-channel), the PPG signal dispersion increases with the dynamics of blood flow. This increase in signal dispersion causes the pulse pressure wave to alternate between maximum and minimum amplitude. If there is a heartbeat, the pulsation amplitude indicates the pulsation of arterial blood.

The DC level (e.g., the mean value of the periodic function) of the PPG signal represents the non-pulsatile component of the blood. The non-pulsating component of the blood includes non-pulsating arterial blood, venous blood, and absorption in background tissues (eg, bone, cartilage, and extra-cellular fluids). .

In the human body, the pulsating component is typically 10% or less of the total PPG signal. This property is not observed in PPG signals where the source of the PPG signal sensed by the PPG sensor is only a reflection from an inert surface. If the PPG signal is a reflection from an inactive surface, the pulsation amplitude may vary irregularly with respect to the DC level of the signal. For example, if the ratio of the AC and DC components of the PPG signal is greater than 0.1 or less than 0.01, the device may determine that the PPG signal obtained is not modulated by blood.

In embodiments where the PPG sensor is multi-channel, the device may be configured to determine the PPG data between the PPG signal generated using the first channel of the PPG sensor and the PPG signal generated using the second channel of the PPG sensor. Covariance can be determined. The device may determine that the PPG data indicates blood flow when it exceeds a predetermined threshold.

At block 230, the device determines whether the PPG data segment is valid. For example, when the device measures the pulsation amplitude, the device may determine whether the pulsation amplitude is within a defined range, corresponding to a valid PPG signal. A valid PPG signal is a PPG signal whose pulsation amplitude is between about 0.01 and about 0.1 (for a single channel PPG sensor).

For example, the device determines whether 0.01 ≦ pulse amplitude ≦ 0.1 is true. In the case of a multi-channel PPG sensor, the device may determine whether the covariance exceeds a threshold. In response to determining that the segment is valid (eg, when .01 ≦ pulse amplitude ≦ 0.1 is true for a single channel PPG sensor or when the covariance exceeds a threshold for a multichannel PPG sensor), 200 continues to block 240. In response to determining that the segment is invalid, the method 200 continues to block 235.

At block 235, the device may reject the PPG data segment and / or generate a notification. For example, the device may reject the segment as described above. In rejecting the segment, for example, the device does not attempt to determine any health indicator using the PPG data segment. The device may also provide a notification indicating that the PPG data segment is not suitable for use in determining health markers. In one or more embodiments, the notification can provide instructions to the user to adjust the device to obtain higher quality PPG data later.

In block 240, the device may determine one or more health markers from the PPG data. Examples of different various health markers that the device can determine from the PPG data include, but are not limited to, heart rate, heart rate variability (HRV), stress, blood pressure, respiration, and arterial tension. For example, the device may determine a PPG RR interval, where "RR" is the interval between successive Rs and R represents the peak of the PPG. The PPG RR interval is an HRV for the user.

As is known, HRV is a physiological phenomenon in which the time interval between heart beats changes. HRV is measured by the change in interval between beats. The device may determine the inverse of the HRV and scale the result to obtain an estimate of heart rate. The device may also determine stress for the user. The device can determine the stress by taking the derivative of the HRV.

The device may determine the blood pressure for the user based on an area under the curve (AUC) of the PPG signal. The device may determine systolic blood pressure for the user based, for example, on the AUC of the PPG signal. Larger AUCs correspond to higher systolic blood pressure.

As an illustrative and non-limiting example, some studies have shown that very low frequency (VLF) fluctuations in systolic blood pressure are associated with PPG, where systolic blood pressure is the PPG amplitude over a 10 minute period. It was found to have a correlation coefficient of approximately -0.81 with the VLF variation of (AM) and a correlation coefficient of approximately 0.83 with the PPG baseline (BL).

At block 245, the device may validate the health indicator (s) determined at block 240. In one or more embodiments, the device validates the health indicator based on previous determinations of the health indicator for the user. In one example, the device may determine whether the health marker determined at block 240 is within a predetermined physiological range for the health marker in humans. Measurements of heart rate can be compared to known heart rate ranges. HRV can be compared with known HRV ranges. As described in more detail herein, the physiological range with which the health metric is compared can be adjusted based on inertial sensor data.

In another example, the device may validate the health indicator by determining whether the health indicator determined at block 240 is within a predetermined amount of health markers of the same type measured immediately prior to the time series of such health markers. Can be. The device may, for example, determine whether the amount of change observed in the health marker in the time series of such health markers is within acceptable levels.

In one or more embodiments, the validation performed at block 245 includes performing post hoc correction on the PPG data segment. In some cases, the device may apply one or more correction techniques to the PPG data segment. Additional aspects regarding validation are described in more detail with reference to FIG. 5.

At block 250, the device may output the health indicator (s) determined at block 240. More specifically, the device may output the health indicator (s) determined at block 240 in response to successful validation of each health indicator.

3 illustrates an example method 300 of quality estimation for the PPG data. The method 300 may be performed by an apparatus as described with reference to FIG. 1 herein. In one or more embodiments, the method 300 may be performed by the apparatus to implement the block 205 of FIG. 2.

At block 305, the device optionally determines a covariance for the PPG data segment. In one or more embodiments, if the PPG sensor is a multi-channel PPG sensor, the device may comprise a PPG signal from one channel of the PPG sensor (eg, a first LED having a first wavelength, such as red). Covariance of a PPG signal from a second channel (eg, a second LED having a second wavelength, such as infrared light) of the PPG sensor may be determined. In certain embodiments, the apparatus may optionally determine the variance as described with reference to the block 205 of FIG. 2.

)에 대해 제공하는 평균 데이터량(예를 들면, 데이터 비트(bits of data))을 나타내는 양이다. 보다 공식적으로, 상호 정보량은 하나의 랜덤 변수가 다른 랜덤 변수에 관해 제공하는 평균 정보량(예를 들면, 비트단위로)으로서 정의된다. 하기 식 1은 식으로 표현된 상호 정보량(MI)을 보여준다.At block 310, the apparatus determines mutual information for the PPG data segment. The amount of mutual information is given by another given measurement x (t +)

) Is an amount representing an average amount of data (for example, bits of data) to be provided. More formally, the amount of mutual information is defined as the average amount of information (eg, in bits) that one random variable provides for another random variable. Equation 1 below shows the mutual information amount (MI) expressed by the equation.

(1)

(One)

)이 주어진 주기 신호에 대해 상대적으로 높을 것이다.

가 커짐에 따라, 비-지연(non-delayed) 측정치 x(t)가 x(t+

)에 관해 제공하는 정보는 적어진다. 상기 상호 정보량의 감소는 상기 지연이 상기 신호의 주기성(상기 신호에 주기성이 있는 경우)과 일치할 때까지 계속된다.Equation 1 represents a high amount of mutual information for a delay corresponding to a periodic signal. Since the two data sets being compared are not very different, the amount of mutual information is

Will be relatively high for a given periodic signal.

As becomes larger, the non-delayed measurement x (t) becomes x (t +

Less information is provided. The decrease in the amount of mutual information continues until the delay matches the periodicity of the signal (if the signal has periodicity).

At block 310, as part of determining the amount of mutual information for the PPG data, the apparatus may determine several different amounts. In one or more embodiments, the apparatus determines a measure of SNR. The apparatus determines, for example, the peak SNR. The apparatus may also determine the width based on the characteristic width of the peaks seen in the information spectrum. As an example, the width may be the full-width half-max (FWHM) of a particular peak. In general, the presence of white noise makes the FWHM smaller because there are more points in the surrounding areas of the peak. Since the block 310 is part of a normalized windowed approach, the peak maximum is lower than the adjacent lags. The apparatus may also determine an estimate of the period of the PPG signals specified by the PPG data.

At block 315, the device may estimate the quality (eg, signal quality) for the PPG data segment. In one or more embodiments, the device may classify the signal quality as low, medium, or high. For example, the apparatus may compare the quality score with predetermined ranges for the quality score, which map to different classes of signal quality. The apparatus may assign a signal quality to the PPG data segment based on which range the quality score belongs to.

In one or more embodiments, the quality score may be a measure of the SNR. In one or more embodiments, the quality score can be determined using measurements of covariance and / or variance and the SNR. As an illustrative and non-limiting example, the device may calculate the quality score by multiplying the covariance and / or variance by a weighting factor and multiplying the measurement of the SNR by another weighting factor and summing the results. In any case, the device estimates the quality of the PPG data segment as low, medium, or high and outputs the estimated quality. As described with reference to FIG. 2, the block 210 may evaluate the signal quality to continue processing the PPG data segment or reject the PPG data segment based on the estimated quality.

The following discussion illustrates an example method of determining the amount of mutual information in a segment of PPG data, as performed by the apparatus and described herein with reference to block 310 of FIG. 3. The device may store the PPG data segment. As discussed, the PPG data segment, also referred to as data array x, is for a specific window. The PPG data may be detrended and filtered PPG samples.

t=

를 포함할 만큼 충분히 크다. 다른 샘플 크기들이 이용될 수 있다. 본 명세서에서 제공되는 상기 예는 제한하고자 한 것이 아니라 예시를 위한 것이다.In one example, the PPG data segment may include 200 samples. The number of samples may be in a range, where the bottom of the range is small enough to be in a respiratory sinus arrhythmia (RSA) cycle to avoid gradual periodic drift, the top of the range being of interest Small enough to detect specific frequency components (eg, heart rate), a

t =

Big enough to contain Other sample sizes may be used. The examples provided herein are for purposes of illustration and not limitation.

만큼 지연된 데이터 어레이를 생성할 수 있으며, 여기서

는 1에서 N까지 변하여 N개의 상이한 (지연된) 데이터 어레이들을 생성한다. 일단 상기 데이터 어레이들이 생성되면, 상기 장치는 X와 X(t+

) 사이에서 상기 상호 정보량을 계산할 수 있다. 상기 장치는 상기 상호 정보량을 다른 데이터 어레이에

가 증가하는 순서로 저장할 수 있다.The device may generate delayed points from the PPG data. In one or more embodiments, the device comprises, from data array x,

Create a data array delayed by

U varies from 1 to N to produce N different (delayed) data arrays. Once the data arrays have been created, the device has X and X (t +

The amount of mutual information can be calculated between The device transfers the mutual information amount to another data array.

Can be stored in increasing order.

To calculate the average amount of mutual information, the apparatus processes a time series array X of length N. The device may move the data vertically to normalize X and start at zero. For example, the apparatus can be divided by the maximum value after subtracting the minimum value, as shown in Equations 2 and 3 below.

(2)

(2)

(3)

(3)

데이터를

만큼 지연시켜

의 결과를 산출할 수 있다.

는

의 지연 이미지(delayed-image)이다. 상기 장치는

를 이용하여 상기 상호 정보량을 계산할 수 있다.The device is

Data

Delayed by

Can yield the result of.

Is

Is a delayed-image of. The device is

The amount of mutual information can be calculated using.

의 2차원 묘사(portrait)이므로, 상기 파티션들(partitions)은 2차원 박스 파티션들일 수 있다. 특정 실시예들에서, 빈 크기는 일정할 수 있고 하기 식 4를 이용하여 계산될 수 있다.In one or more embodiments, the apparatus can divide the distribution functions into a predetermined number of bins. In this example

The partitions may be two-dimensional box partitions because of the two-dimensional portrait of. In certain embodiments, bin size may be constant and may be calculated using Equation 4 below.

(4)

(4)

및

의 길이이다. 상기 빈들이 너무 큰 경우, 상기 분포는 빈당(per bin) 더 많은 포인트들을 가지며 정확성이 향상된다. 상기 빈들이 너무 크기 때문에, 상기 장치의 상기 프로세서는 짧은 거리들(시간 래그들)에 걸친 상호 정보량의 변화를 추적하지 못 할 수 있다. 상기 빈 크기가 보다 작은 경우, 정확성은 저하되지만, 전체적인 해상도는 증가한다.Based on Equation 4, there are N ² boxes in this phase portrait. In the example of equation 4, N is

And

Is the length of. If the bins are too large, the distribution has more points per bin and the accuracy is improved. Because the bins are so large, the processor of the device may not be able to track changes in the amount of mutual information over short distances (time lags). If the bin size is smaller, the accuracy is degraded, but the overall resolution is increased.

In one or more embodiments, the number of bins may be dynamically adapted in operation. The apparatus may adapt the number of bins such that the number of bins is not constant for regions of the phase space.

In certain embodiments, at this point, the device may check for variance. In practice, the variance will not be zero. However, if the variance is calculated as zero, then the mutual information amount will also be zero.

의 모든 인스턴스에 대해 상기 박스 검색 변수들을 순회할(traversing) 수 있다. 특정 실시예들에서, 상기 장치는 구현 시 중첩 for 루프(nested for-loop)를 이용하여 상기 순회(traversal)를 수행할 수 있다. The apparatus uses box-search variables s ₁ and s ₂ , where a particular box corresponds to point (s _1i , s _2j ). The device is

It is possible to traverse the box search variables for all instances of. In certain embodiments, the apparatus can perform the traversal using a nested for-loop in implementation.

의 확률 분포를 의미하며, p_y는

의 확률 분포를 의미한다. 상기 장치는, 식 5 내지 식 14에 의해 정의되는 조건들을 만족시키는,

의 모든 인스턴스들을 구할 수 있다.For example, s ₂ may be a partition along the inner-horizontal axis of the phase space, but s ₂ is a partition along the inner-vertical axis of the phase space. Referring to Equations 5 to 14, p denotes a joint probability distribution, and p _x

Means the probability distribution of, and p _y is

Means the probability distribution. The apparatus satisfies the conditions defined by equations 5 to 14.

All instances of are available.

(5)

(5)

(6)

(6)

(7)

(7)

(8)

(8)

(9)

(9)

(10)

10

(11)

(11)

(12)

(12)

(13)

(13)

(14)

(14)

가 증가함에 따라 종말 포인트들(end points)이 제거된다. 이와 같이, N-

아규먼트(argument)는 상기 식 12 내지 식 14에서 정규화 분모(normalizing denominator)에 이용된다.In this example, because the process requires pairs of values, data is lost in the technique described above. By time-delaying the data,

As points increase, end points are eliminated. Thus, N-

Arguments are used for normalizing denominators in Equations 12-14.

The apparatus may determine the final amount of mutual information given the probabilities p, p _x , and p _y using Equation 15, which is the same as Equation 1 below.

(15)

(15)

의 모든 값에 대해 반복될 수 있다. 상기 장치는 새로운 어레이에 대해 피크 검출을 더 수행할 수 있다. 특히, 상기 장치는 심박수에 상응하는 로컬 피크들을 검출할 수 있다. 일반적으로, 상기 로컬 피크들은, 상기 PPG 데이터가 얻어지는 상기 사용자의 심박수를 제공하는 주파수 변환(frequency transform)을 갖는 델타

t에 상응하는,

의 값들에 상응할 수 있다.The calculations are performed to obtain a distribution of mutual information amount over all delays,

Can be repeated for all values of. The apparatus may further perform peak detection on the new array. In particular, the device can detect local peaks corresponding to heart rate. In general, the local peaks have a delta with a frequency transform that provides the heart rate of the user from which the PPG data is obtained.

corresponding to t,

May correspond to the values of.

)가 존재한다.The apparatus may repeat the operations three times with the mutual information amount bins being added to each previous bin. Once the mutual information quantity output array is filled with the output of four separate and consecutive mutual information quantity calculations, the apparatus removes a first mutual information quantity output array from the mutual information quantity output array. More specifically, MI ^* (calculated as shown in equation 17)

) Exists.

(17)

(17)

)는 특정 시간 지연(래그)에서 상호 정보량의 완결된 값이다. 다음으로 상기 장치는 3개의 상이한 윈도우들로부터의 나머지 3개의 값들에 완결된 MI_i(

) 포인트들을 추가할 수 있다. 이 프로세스는, LED들(예를 들면, PPG 센서)의 비활성화(disabling) 이전에, 4개의 상호 정보량 빈들 모두가 채워질 수 있도록 한다. 상기 상호 정보량 빈들이 채워지지 않은 경우, 상기 상호 정보량은 손가락 배치에 더 민감하다.Each MI _i (

) Is the completed value of the mutual information amount at a specific time delay (lag). The device then completes MI _i (completed with the remaining three values from three different windows).

) Points can be added. This process allows all four mutual information bins to be filled before disabling the LEDs (eg, PPG sensor). If the mutual information amount bins are not filled, the mutual information amount is more sensitive to finger placement.

The apparatus may repeat the process except that the mutual information quantity output array is a sliding window of mutual information quantity (for example, referred to as MI ^* above). For example, after the device removes MI ₃ from MI ^* , MIi is moved to MI _{(i + 1)} , without MI ₀ . The device adds the latest calculation of MI to this empty buffer point and then recalculates MI ^* .

)에 관한 정보를 거의 제공하지 못함을 암시한다.The device can monitor the relative change in MI (at the peak corresponding to heart rate). In one or more embodiments, if the amount of mutual information falls too low, eg, below a threshold level of mutual information, the device is no longer periodic or sufficient in periodicity to the PPG data. Decide not. This means knowing X is X (t +

Imply little information about).

4 illustrates an example method 400 of correcting and filtering the PPG data. The method 400 may be performed by an apparatus as described with reference to FIG. 1 herein. In one or more embodiments, the method 400 may be performed by the apparatus to implement the block 220 of FIG. 2.

At block 405, the apparatus determines whether the quality of the segment is high. For example, the apparatus may classify the PPG data segment as described with reference to the block 205 of FIG. 2 and the block 315 of FIG. 3. In general, the apparatus may apply dynamic processing to the PPG data segment based on the estimated quality.

In the example of FIG. 4, the device applies corrective processing to the PPG data segment in response to determining that the quality is medium (eg, not high). Recall from previous discussions of FIGS. 2 and 3 that low quality segments of PPG data are rejected at block 215, the apparatus deals with PPG data classified as high or medium at block 405.

In response to determining that the quality of the PPG data segment is high, the method 400 continues to block 415. In this case, the device may omit the correction process for the PPG data segment. In response to determining that the quality of the PPG data segment is not high, e.g., intermediate, the method 400 continues to block 410, wherein the apparatus performs a correction process on the PPG data segment. .

In block 410, the device may perform a correction process on the PPG data segment. In one or more embodiments, the apparatus removes noise from the segment. If the PPG data segment is of intermediate quality, the segment may be corrupted with different noise components (eg, artifacts in motion) that are not inherent parts of the basic PPG morphology.

In one or more embodiments, as part of the block 410, the device may apply a filter to remove high frequency components of the PPG data, which is not related to heart rate analysis. For example, the device may filter the PPG data segment using a fifth order IIR filter. Since the derivative peaks are more pronounced than normal PPG data, the above described process can also be performed with the derivatives of the time of the filtered PPG data.

In one or more embodiments, the apparatus may perform the conversion (eg, frequency conversion) of the PPG data only in response to determining that the PPG data segment contains sufficient noise to prevent unnecessary calculations. have. As an illustrative and non-limiting example, the apparatus may use variance to determine whether the PPG data includes noise. The device may calculate the variance using a data set of accelerometer data. Accelerometer data is associated with increased motion artifacts. The device may determine the variance of the data set x, where x is an array of accelerometer values synchronized in time with the PPG data segment.

The device may parse the array of accelerometer values into smaller segments. Each segment may contain 50 values, which the device determines variance. For each window, the device determines the variance and determines whether the calculated variance is greater than the threshold σ, where σ = 0.3. In response to the variance exceeding the threshold, the device determines that the PPG data segment contains noise. If the variance does not exceed the threshold, the device determines that the PPG data segment does not contain noise, and performs some additional correction processing and transforms the PPG data into the frequency domain using a transform. The step of performing a translation can be omitted. It is to be understood that the specific values provided above are for purposes of illustration and not limitation.

In certain embodiments, when noise is detected using the accelerometer data, the device may decompose the signal to filter certain components. Since the process described above is performed in near real time, the apparatus can decompose PPG data segments, which will occur within windows of a predetermined amount of time and / or number of samples. As an illustrative and non-limiting example, the number of samples can be 800.

At block 410, the device may transform the PPG data using a continuous complex wavelet transformation function. In certain embodiments, the device uses a Morlet wavelet to represent the PPG data. The apparatus may generate a first window of N samples, constructed such that a point with detected noise (eg, sigma> 0.3) is located in the center of the window. The edge-coefficient effect is mitigated by the zero-padding of the wavelet function, but the instrument can locate the point of interest in the center of the wavelet window. The Morlet wavelet is given by Equations 18 and 19 below.

(18)(19)

(18) (19)

및

상에서 각각 계산된다.In the examples of equations 18 and 19, s is referred to as dilation and is inversely related to the frequency being analyzed. In the example of FIG. 4, s is an array of values from lowest frequency s ₀ = 2 * dt to higher multiples of s ₀ . The wavelet transform (scaled version of the Morlet mother) for each daughter wave is a discrete Fourier transform

And

Is calculated for each phase.

(20)

20

(21)

(21)

Referring to Equation 21, W _n represents an array of wavelet coefficients for a specific time index n, where s is an independent variable inside the convolution. Where s is the scale of the wavelet and is approximately a frequency term. Each W _n is a set (corresponding to scale) of wavelet coefficients at a particular time index. Equations 20 and 21 evoke the convolution theorem by calculating the inverse Fourier transform of the product shown in equation 21. Equation 20 represents a Discrete Fourier Transform (DFT) of the signal window from the PPG data.

The signal p 'window exists in a two-dimensional matrix, with rows representing the temporal index for a particular scale (s) and columns representing the real and imaginary components of the transform. The apparatus can convert the two-dimensional matrix to a scalar value by calculating the complex modulus of the complex elements.

The device may calculate the magnitude of the wavelet transform from the scale band corresponding to the expected signal location. For this assay, s = 36 is the appropriate size to capture a particular physiological signal. The device can compare this size with the size of the remaining wavelet image. Since the suspected noise is already located in the center of the window, a large magnitude is expected at this center point.

The apparatus may also attenuating elements of the transform that are significantly larger than the average scale band size. In one or more embodiments, to avoid filtering of artifacts in inverse transform, the apparatus uses a smooth blending function defined as a quadratic function with zero-crossing at each end of the noise window. Can attenuate the elements. As a result, the device applies a minimum attenuation at the end points of the window (smoothly blended with the side band spectrum) and a maximum attenuation at the center of the noisy portion.

Equation 22 illustrates the filtering of the wavelet transform coefficients W _n as a function of time index n. As such, equation 22 represents the attenuated signal as the wavelet transform coefficients, which are then used for subsequent inverse transform to recover the filtered time series.

(22)

(22)

The device may also reconstruct the PPG signal. In this example, orthogonal transformations are not assumed. As such, the apparatus may use the delta function to determine the inverse transform and sum the real components of W _n ^* into appropriate de-scaling coefficients.

Within the present disclosure, wavelet transform is used for illustrative purposes. The configurations of the present invention are not limited by the examples provided above. In other embodiments, for example, different transforms may be used. For example, other windowed frequency domain transforms may be used instead of wavelets.

At block 415, the apparatus is based on the periodicity of the signal determined from mutual information amount operations as described with reference to block 205 of FIG. 2 and block 310 of FIG. 3. Determines whether the frequency of is high. The device compares the frequency of the signal with a frequency threshold. In one or more embodiments, the frequency threshold may be set to a heart rate of 100 beats per minute (BPM), corresponding to a tachycardic signal. If the PPG signal is prone to tachycardia (eg, heart rate exceeds 100 BPM), a suitably tuned comb filter can be used for heart rate estimation. For PPG signals that are not tachycardia, such signals can be analyzed using another approach, such as when applying a band pass filter.

Thus, in response to determining that the frequency of the PPG data is low (eg, not exceeding the frequency threshold), the method 400 continues to block 420. At block 420, the device applies a band pass filter to the PPG data segment. In response to determining that the frequency of the PPG data is high (eg, exceeding the frequency threshold), the method 400 continues to block 425. At block 425, the device applies a comb filter to the PPG data.

5 illustrates an example method 500 of validating health indicator (s). The method 500 may be performed by an apparatus as described with reference to FIG. 1 herein. In one or more embodiments, the method 500 may be performed by the apparatus to implement the block 245 of FIG. 2. In certain embodiments, the device may further perform post-correction of the PPG data.

In block 505, the device may validate the health indicator determined in block 240 of FIG. 2. In one or more embodiments, the device may compare a current health indicator (eg, the health indicator determined in block 240) with recent historical measurements of the same type of health indicator. As an illustrative and non-limiting example, the device may evaluate the most recent N estimates of heart rate for the user when N is an integer value. In this example, the time series of health markers has 10 data items, which are 89 90 89 52 213 212 211 211 92 92 (in BPM). The above example illustrates a possible error, in which a fluctuation in heart rate is first seen when the heart rate estimate transitions from 89 to 52. In addition, a possible error is illustrated when the estimated heart rate increases significantly from 52 to 213, which continues until it stabilizes again from 92th to 92.

Thus, in one or more embodiments, the device may determine whether the health indicator values in the time series of estimated heart rate values fall above or below the upper limit. The upper and lower limits may be known physiological limits for an individual of the same or similar health as the user. The device may discard any health indicator that is above the upper threshold or below the lower threshold. For example, in the example time series of heart rates, values such as 52, 213, 212, and 211 may be discarded.

In one or more embodiments, the device may determine whether the amount of change between two consecutive health markers in the time series of health markers exceeds a threshold change amount. If so, the device may discard the health marker (s) that are considered inaccurate. Referring to the example data set of heart rates, the device may discard 52 heart rate and 213 heart rate, where the change from the previous heart rate is too large.

In one or more embodiments, the device may use inertial sensors to validate health indicators for the user. The device may determine the activity level of the user based on inertial sensor data, such as, for example, gyroscope sensor data, accelerometer sensor data, and / or location (place) data over time. . The inertia data may be correlated with health indicators such as heart rate and HRV, etc., for the user on a historical basis. Thus, the inertial sensors can be used to indicate the likelihood that noise is present in the PPG data, as described herein with reference to FIG. 4 (eg, the block 410). The health markers determined at block 240 can be used to set an expected range of health markers that can be compared for validation.

For example, the device may determine the activity level for the user for a given point in time based on inertial sensor data. The device may determine a heart rate range for the user given the activity level. The device may compare the heart rate determined in block 240 with the heart rate range. In response to determining that the estimated heart rate is within the heart rate range, the device may determine that the heart rate of block 240 is valid. In response to determining that the heart rate determined at block 240 is above or below the heart rate range, the device may determine that the estimated heart rate is invalid.

At block 510, the device determines whether the health indicator is valid as described. In response to determining that the health indicator is valid, the method 500 continues to block 515. In block 515, the device outputs the health indicator. For example, the device may output the health indicator to a display and / or store the health indicator in a memory. In response to determining that the health indicator is invalid, the method 500 proceeds to block 520.

At block 520, the device may perform RR correction on the PPG data segment. For example, if it is determined that the health indicator is invalid, the device may perform corrective action to obtain a valid health indicator. As an illustrative and non-limiting example, if the device determines that the health marker (eg, heart rate) has increased approximately twofold from previous measurements of the health marker, the device may remove the peak. Referring to the previous example of estimated heart rates, the increase in heart rate estimates from approximately 92 to approximately 213 may have been the result of detecting an unrelated peak in the time series data. By eliminating the peak, the estimated heart rate can be returned to within acceptable tolerance.

At block 525, the device may perform validation on the RR corrected PPG data segment. The apparatus may substantially perform the block 525 as described with reference to the block 505. At block 530, the apparatus determines whether the RR corrected PPG data segment is valid. If so, the method 500 may loop back to the block 520 to output the health indicator using the RR corrected PPG data segment. If not, the method continues to block 535.

At block 535, the device may reject the RR corrected PPG data segment and / or generate a notification. For example, the device may delete the RR corrected PPG data segment. In rejecting the segment, for example, the device does not attempt to determine any health markers using the RR corrected PPG data segment. The device may also provide a notification indicating that the PPG data segment is not suitable for use in determining health markers. In one or more embodiments, the notification can provide instructions to the user to adjust the device to obtain higher quality PPG data.

As discussed, in response to rejection of the PPG data segment, the device may generate a notification. In one or more embodiments, the apparatus may customize the provided instructions based on the particular aspect or dimension that "failed" resulting in rejection of the PPG data segment. For example, the notification may indicate a particular test that failed or a particular aspect of the PPG data segment that is considered insufficient. The device may also provide instructions for adjusting the wearable device, for example specific to the particular reason for which the PPG data segment was determined to be insufficient quality.

As an illustrative and non-limiting example, in response to determining that the relative pulsation amplitude described with reference to block 225 represents a non-biological contact, the device may generate a notification to the user. . The notification may be a push notification indicating that no health indicators have been detected. In addition, in response to the detection of the non-biological contact, the device may be powered off. The power off may be triggered after a predetermined amount of time after detecting the non-biological contact or in response to the detection of the non-biological contact for a predetermined or minimum amount of time. The power off can preserve battery life and prevent any further logging of data. The notification may also inform the user of the power off. For example, the device may display a countdown to the power off.

As another illustrative and non-limiting example, in response to determining that the wavelet processing described with reference to block 410 represents a step response in the PPG data, the device generates a notification to the user. can do. The notification may be a push notification. The type of noise detected at block 410 may be due to a loose contact between the device and the user. Thus, the notification may instruct the user to strengthen the mechanism for associating the device with the user. In the case of a smart watch, for example, the notification may instruct the user to tighten the wristband of the watch.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. Nevertheless, several definitions that apply throughout this document will now be presented.

As used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise.

The term “approximately”, as defined herein, means almost correct or exact and means close but not precise in value or amount. For example, the term “approximately” can mean that the described characteristic, parameter, or value is within a predetermined amount of an exact characteristic, parameter, or value.

The terms "at least one", "one or more", and "and / or" as defined herein, unless expressly stated otherwise, Conjunctive and disjunctive open-ended expressions in operation. For example, the expressions “at least one of A, B, and C”, “at least one of A, B, or C”, “one or more of A, B, and C”, and “A, B, and / Or C ″ means A alone, B alone, C alone, A and B together, A and C together, B and C together, or A, B and C together.

The term “automatically”, as defined herein, means no user intervention.

The term “computer readable storage medium,” as defined herein, means a storage medium that contains or stores program code for use by or in association with an instruction execution system, an apparatus, or an apparatus. As defined herein, a “computer readable storage medium” is not a transient, propagating signal by itself. The computer readable storage medium may be, but is not limited to, an electronic storage device, magnetic storage device, optical storage device, electromagnetic storage device, semiconductor storage device, or any suitable combination thereof.

Various forms of memory, as described herein, are examples of computer readable storage media. A non-limiting list of more specific examples of computer readable storage media includes portable computer diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM, or flash memory). , Static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, or floppy disk And the like.

As defined herein, the term "if" (if) is, depending on the context, in response to "when" or "upon" or "upon". (in response to) "or" responsive to. " Thus, the phrase "if it is determined" or "if the [referred condition or event] is detected" means "on decision" or "in response to a decision" or "in response to a decision" or "a condition or event mentioned" depending on the context. Or "in response to" [detected condition or event] or "responsive to detection of [referred condition or event]".

As defined herein, the term "responsive to" and similar languages as described above, such as "if" (if), "when" "Or" upon "means an easy response or response to an action or event. The response or reaction is performed automatically. Thus, if the second act is "sensitively" to the first act, then there is a causal relationship between the occurrence of the first act and the occurrence of the second act. The term "sensitive to" refers to the causal relationship.

As defined herein, the terms "one embodiment", "an embodiment", "one or more embodiments", "particular embodiments" Or similar language means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment described within the present disclosure. Thus, throughout the present disclosure the phrase forms “in one embodiment”, “in an embodiment”, “in one or more embodiments”, “specific "Particular embodiments", and similar language, may all mean the same embodiment, but are not necessarily so. The terms "embodiment" and "arrangement" are used interchangeably within this disclosure.

The term "output", as defined herein, refers to a physical memory element, for example, storing on a device, writing to a display or other peripheral output device, sending to or transmitting to another system. , Or exporting.

The term “processor”, as defined herein, means at least one hardware circuit. The hardware circuitry may be configured to perform instructions contained in program code. The hardware circuit may be an integrated circuit. Examples of processors include central processing units (CPUs), array processors, vector processors, digital signal processors (DSPs), field-programmable gate arrays (FPGAs), programmable logic An array of programmable logic arrays (PLAs), application specific integrated circuits (ASICs), programmable logic circuits, and controllers.

As defined herein, the term “real time” refers to a process response that senses that the user or system is immediately instantaneous about a particular process or decision to make, or that allows the processor to catch up with any external process. The level of processing responsiveness.

The term “substantially”, as defined herein, does not have to be precisely achieved for a characteristic, parameter, or value described, for example, tolerances, measurement errors, measurements Deviations or variations, including measurement accuracy limitations, and other factors known to those skilled in the art, do not exclude the effect that the feature is intended to provide. , It can happen.

The term "user", as defined herein, means a human.

The terms first, second, etc. may be used to describe various elements herein. These terms are to be used only to distinguish one element from another, unless stated otherwise or otherwise clearly stated in context, and such elements should not be limited by these terms.

The computer program product may include computer readable storage medium (s) having computer readable program instructions for causing a processor to perform aspects of the present invention. Within this specification, the term "program code" is used interchangeably with the term "computer readable program instructions."

The computer readable program instructions described herein may be executed from a computer readable storage medium to each computing / processing device, or to a network, such as the Internet, a local area network (LAN), a wide area network. Network (WAN), and / or can be downloaded to an external computer or external storage device via a wireless network. The network may include edge devices including copper transmission cables, optical transmission fibers, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface at each computing / processing device receives computer readable program instructions from the network and conveys the computer readable program instructions for storage on a computer readable storage medium in each computing / processing device. .

Computer-readable program instructions for performing operations on the inventive constructions described herein include assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, It may be source code or object code written in any combination of one or more programming languages, including microcode, firmware instructions, or object-oriented programming language and / or procedural programming language. Computer-readable program instructions may specify state-setting data. The computer readable program instructions are entirely at the user's computer, partly at the user's computer, as stand-alone software packages, partly at the user's computer and partly at the remote computer, or entirely at the remote computer or server. Can be executed. In the latter scenario, the remote computer can be connected to the user's computer via any type of network, including a LAN or WAN, or a connection to an external computer can be made (eg using an Internet service provider). Through the internet). In some cases, electronic circuitry, including, for example, programmable logic circuits, FPGAs, or PLAs, may be used to personalize the electronic circuitry to perform aspects of the inventive arrangements described herein. By using the status information of the enabled program instructions, the computer readable program instructions can be executed.

Certain aspects of the configurations of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products. It will be appreciated that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer readable program instructions, eg, program code. .

Such computer readable program instructions may be provided to a processor of a computer, special purpose computer or other programmable data processing device to produce a machine such that the instructions executed through the processor of the computer or other programmable data processing device are Flowcharts and / or block diagrams may be created to create means for implementing the functions / acts specified in the block or blocks. In this way, by operatively combining the processor and program code instructions, the machine of the processor is converted to a special purpose machine for performing the instructions of the program code. Such computer readable program instructions may also be stored on, and stored on, a computer readable storage medium that can direct a computer, programmable data processing device, and / or other devices to function in a particular manner. The instructions may be adapted to include an article of manufacture, including instructions that implement aspects of the operations specified in the flowchart and / or block diagram block or blocks.

The computer readable program instructions may also be loaded on a computer, other programmable data processing device, or other device such that a series of operations to be performed on the computer, other programmable device or other device creates a computer implemented process, such that The instructions executed on a computer, other programmable device, or other device may cause the functions / acts specified in the flowchart and / or block diagram block or blocks to be implemented.

Flow diagrams and block diagrams in the drawings illustrate the architecture, function, and operation of possible implementations of systems, methods, and computer program products in accordance with various aspects of the configurations of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, including one or more executable instructions for implementing particular operations. In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, depending on the function involved, two blocks shown in succession may be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order. Each block of the block diagrams and / or flow chart examples, and combinations of blocks in the block diagrams and / or flow chart examples, perform specific functions or acts or perform combinations of special purpose hardware and computer instructions, It will also be appreciated that it may be implemented by special purpose hardware based systems.

Corresponding structures, materials, acts, and equivalents of all the elements or step plus function elements that can be found in the claims below, in combination with other claimed elements, as specifically claimed It is intended to include any structure, material, or action for carrying out this function.

The description of the embodiments provided herein is for illustration only and is not intended to be limited or limited to the disclosed form and examples. The terminology used herein is intended to explain the principles of the constructions of the invention, the practical application or technical improvement of techniques found in the marketplace, and / or having ordinary knowledge in the art to which the invention pertains. It was chosen to enable those skilled in the art to understand the embodiments disclosed herein. Modifications and variations may be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described inventive configurations. Accordingly, reference should be made to the following claims, rather than the foregoing disclosure, as indicating the scope of such features and implementations.

Claims (24)

Determining a quality estimate for a segment of PPG data;
In response to determining that the quality estimate exceeds a quality threshold, using a processor to filter the segment of the PPG data based on an estimate of the periodicity of the segment of the PPG data;
Using the processor, determining a health marker for the segment of the PPG data;
Validating the health marker based on prior determination of the health marker from PPG data; And
In response to the validation, outputting the health indicator. The method of claim 1, wherein the PPG data is multi-channel PPG data, and the determining of the quality estimate comprises:
Determining a covariance between PPG data corresponding to the first channel of the multi-channel PPG data and PPG data corresponding to the second channel of the multi-channel PPG data. The method of claim 1, wherein determining the quality estimate is:
Determining mutual information for the segment of the PPG data. The method of claim 1, wherein the filtering step:
Selecting a filter from a plurality of filters based on the estimate of the periodicity of the segment of the PPG data; And
Applying the selected filter to the PPG data. The method of claim 1, wherein validating the health indicator comprises:
Determining whether the health marker is within a predetermined physiological range for the health marker. The method of claim 1, wherein validating the health indicator comprises:
Determining whether the health marker has changed by more than a predetermined amount compared to a previous determination of the health marker. The method of claim 1, wherein validating the health indicator comprises:
Performing RR correction of the segment of the PPG data. According to claim 1,
De-noising the segment of the PPG data prior to the filtering step. The method of claim 8,
Determining whether to perform the denoising based on inertial sensor data. According to claim 1,
Determining a pulsatile amplitude of the segment of the PPG data; And
Determining whether the segment of PPG data is valid based on the pulsation amplitude. The method of claim 1, wherein the health indicator is heart rate. The method of claim 1, wherein the health indicator is heart rate variability. The method of claim 1, wherein the health indicator is stress. A memory configured to store instructions; And
A processor coupled to the memory and configured to initiate operations in response to execution of the instructions, the operations include:
Determining a quality estimate for a segment of PPG data;
In response to determining that the quality estimate exceeds a quality threshold, filtering the segment of the PPG data based on an estimate of the periodicity of the segment of the PPG data;
Determining a health indicator for the segment of the PPG data;
Validating the health marker based on previous determination of the health marker from PPG data; And
In response to the validation, outputting the health indicator. 15. The method of claim 14, wherein the PPG data is multichannel PPG data, and the determining of the quality estimate comprises:
Determining a covariance between the PPG data corresponding to the first channel of the multichannel PPG data and the PPG data corresponding to the second channel of the multichannel PPG data. 15. The method of claim 14, wherein determining the quality estimate comprises:
Determining the amount of mutual information for the segment of the PPG data. The method of claim 14, wherein the filtering step:
Selecting a filter from a plurality of filters based on the estimate of the periodicity of the segment of the PPG data; And
Applying the selected filter to the PPG data. The method of claim 14, wherein validating the health indicator comprises:
Determining whether the health indicator is within a predetermined physiological range for the health indicator. The method of claim 14, wherein validating the health indicator comprises:
Determining whether the health marker has changed by more than a predetermined amount compared to a previous determination of the health marker. The method of claim 14, wherein validating the health indicator comprises:
Performing RR correction of the segment of the PPG data. The method of claim 14, wherein the processor is configured to initiate executable operations, wherein the operations are:
Denosing the segment of the PPG data prior to the filtering step. The processor of claim 21, wherein the processor is configured to initiate executable operations, wherein the operations are:
Determining whether to perform the denoising based on inertial sensor data. The method of claim 14, wherein the processor is configured to initiate executable operations, wherein the operations are:
Determining a pulsation amplitude of the segment of the PPG data; And
Determining whether the segment of the PPG data is valid based on the pulsation amplitude. A computer program product comprising one or more readable storage media having stored thereon a program, wherein the program performs the method of any one of claims 1 to 13 when executed by at least one processor. .

Images